-
Notifications
You must be signed in to change notification settings - Fork 2.3k
New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
update default net to nn-4f56ecfca5b7.nnue #3816
Conversation
New net trained with nnue-pytorch, started from a master net on a data set of Leela (T60.binpack+T74.binpck) Stockfish data (wrongIsRight_nodes5000pv2.binpack), and Michael Babigian's conversion of T60 Leela data (including TB7 rescoring) (farseer.binpack) available as a single interleaved binpack: https://drive.google.com/file/d/1_sQoWBl31WAxNXma2v45004CIVltytP8/view?usp=sharing The nnue-pytorch branch used is https://github.com/vondele/nnue-pytorch/tree/wdl passed STC: https://tests.stockfishchess.org/tests/view/61a3cc729f0c43dae1c71f1b LLR: 2.95 (-2.94,2.94) <0.00,2.50> Total: 49152 W: 12842 L: 12544 D: 23766 Ptnml(0-2): 154, 5542, 12904, 5804, 172 passed LTC: https://tests.stockfishchess.org/tests/view/61a43c6260afd064f2d724f1 LLR: 2.96 (-2.94,2.94) <0.50,3.00> Total: 25528 W: 6676 L: 6425 D: 12427 Ptnml(0-2): 9, 2593, 7315, 2832, 15 Bench: 6885242
Merged via e4a0c6c, congrats :-) |
Which version of the nnue-pytorch trainer should be considered the most "official", in your opinion? I see that the ReadMe.md file of official-stockfish still points to the nodchip repository, and as I understand it this is no longer the recommended trainer, so it would be nice to point to the best nnue-pytorch trainer at this point. |
https://github.com/glinscott/nnue-pytorch/ is the official trainer... I'll prepare a PR to merge my wdl as soon as get some time. I agree we should update the readme.md file, giving credit to the original nodchip trainer, while pointing to the current official. |
this intends to skip (or more correctly weight) data points that are possibly incorrectly evaluated, i.e. retain data that are more likely to be correct. Was used to train two recent SF nets: official-stockfish/Stockfish#3816 official-stockfish/Stockfish#3808 --no-wld-fen-skipping option can be used to disable the default
this intends to skip (or more correctly weight) data points that are possibly incorrectly evaluated, i.e. retain data that are more likely to be correct. Was used to train two recent SF nets: official-stockfish/Stockfish#3816 official-stockfish/Stockfish#3808 --no-wld-fen-skipping option can be used to disable the default
this intends to skip (or more correctly weight) data points that are possibly incorrectly evaluated, i.e. retain data that are more likely to be correct. Was used to train two recent SF nets: official-stockfish/Stockfish#3816 official-stockfish/Stockfish#3808 --no-wld-fen-skipping option can be used to disable the default
closes official-stockfish#3822 also adjusts readme as requested in official-stockfish#3816 No functional change
the readme has been adjusted |
New net trained with nnue-pytorch, started from a master net on a data set of Leela
(T60.binpack+T74.binpck) Stockfish data (wrongIsRight_nodes5000pv2.binpack), and
Michael Babigian's conversion of T60 Leela data (including TB7 rescoring) (farseer.binpack)
available as a single interleaved binpack:
https://drive.google.com/file/d/1_sQoWBl31WAxNXma2v45004CIVltytP8/view?usp=sharing
The nnue-pytorch branch used is https://github.com/vondele/nnue-pytorch/tree/wdl
passed STC:
https://tests.stockfishchess.org/tests/view/61a3cc729f0c43dae1c71f1b
LLR: 2.95 (-2.94,2.94) <0.00,2.50>
Total: 49152 W: 12842 L: 12544 D: 23766
Ptnml(0-2): 154, 5542, 12904, 5804, 172
passed LTC:
https://tests.stockfishchess.org/tests/view/61a43c6260afd064f2d724f1
LLR: 2.96 (-2.94,2.94) <0.50,3.00>
Total: 25528 W: 6676 L: 6425 D: 12427
Ptnml(0-2): 9, 2593, 7315, 2832, 15
Bench: 6885242